CD4 T cell diversification in the tissue

(1)

CD4 T cell diversification in the tissue

Inauguraldissertation

zur

Erlangung der Würde eines Doktors der Philosophie

vorgelegt der

Philosophisch-Naturwissenschaftlichen Fakultät der Universität Basel

von

Nivedya Swarnalekha

von India 2021

Originaldokument gespeichert auf dem Dokumentenserver der

Universität Basel https://edoc.unibas.ch

(2)

Genehmigt von der Philosophisch-Naturwissenschaftlichen Fakultät auf Antrag von

(Prof.Dr.Gennaro de Libero, Prof.C.G.King, Prof.R.Tussiwand) Basel, 02/03/2021

Dean of Faculty

Prof. Dr. Marcel Mayor

(3)

Table of Contents

Acknowledgements………...………01

Abbreviations...……….02

Summary………...……….04

1. Introduction...05

1.1 Immunology and Vaccines...05

1.2 CD4 T cell memory: friend or foe? ...06

1.2.1 CD4 T cell differentiation...06

1.2.2 CD4 T cell response...06

1.2.3 Signals for Th1/TFH generation, function and memory………....……...…07

2. Immunological memory in barrier tissues...11

2.1 TRM... ...11

2.1.1 Identification...11

2.1.2 Circulation ability...11

2.1.3 Phenotype...12

2.1.4 Factors responsible for TRM generation and maintenance………...…………...……..…………12

2.1.5 TRM function: protective and pathogenic roles……….……….13

2.2 Influenza and CD4 TRM...15

2.2.1 Influenza virus...15

2.2.2 IAV proteins as immune targets………..……….15

2.2.3 Open questions and goals of thesis………..………..16

3. Aim...16

4. Results ...17

Abstract...17

Introduction...17

Results...18

Discussion ...27

Materials and Methods...28

Supplementary Materials...34

5. Discussion...50

5.1 Heterogeneity and tissue-specific genes……….……50

5.2 HIF-1α and possible role in TRM regulation………...………51

5.3 Dynamics of long-term T-B interactions in the tissue………53

5.4 Caveats and Outlook...55

5.4.1 Caveats of TRM study design…………...………55

5.4.2 CD8 T cell depletion to study CD4 TRM in influenza models………..…………56

5.4.3 Contribution of CD4 TRM to OAS………...…….………..56

5.4.4 Relevance of TRM study in the fight against pandemic………56

6. References...57

7. Appendix...61

(4)

(5)

Acknowledgements

Days melted into each other and it is hard to believe many years have passed and my PhD journey is coming to an end. This was a marvelous adventure of learning and discovery both scientifically and personally and I am grateful to the many that have helped me reach the

‘official’ finish line (although there is no real end to learning or discovery).

Words don’t suffice to thank my supervisor, Carolyn King who has been my guru (teacher) and my sakhi (friend) throughout. Thank you for the opportunity. Next, I would like to thank my labmates Marco, Tamara, David, Ludivine, Clemens, Sanketh. Thank you for being my anchor and tailwind. I also give thanks to my Committee members: Professors Roxane Tussiwand, Gennaro de Libero, Dietmar Zehn, and the PhD student community.

I would not be where I am today without the wonderful women in my life who have epitomized beauty, grace, courage and resilience and stood true role models. My late grandmother Jagathambal, my mother Swarnalekha, my mother-in-law Premalatha, my sister-in-law Archana, my aunt Karuna and my best friends: Shilpa, Meha, Aishwarya and Lekshmi. Thank you for the nudge, push, shove and slap.

To all my family, teachers and friends in India and the rest of the world who were instrumental in shaping me into the person I am today, my heartfelt thank you. Special thanks to my parents, Swarnalekha and Mariyappan, my brother Hariprasad, sister-in-law Archana and my nieces Nakshatra and Shanaya.

Last but not the least I would like to thank my better half Sajeev who has stuck with me through thick and thin. Thank you for marrying me at the start of my PhD and still being around (തീമഴ,പൂമഴ! = in rains of fire or flowers)

(6)

Abbreviations

M.tb Mycobacterium tuberculosis HIV-1 Human Immuno Virus -1 CD Cluster of Differentiation TFH T follicular helper

GC Germinal center

APC Antigen presenting cell

MHC Major Histocompatibility Complex TCR T cell receptor

Th1 T helper 1

IFN-γ Interferon gamma

CXCR C-X-C motif chemokine receptor type CXCL C-X-C motif chemokine ligand SLO Secondary lymphoid organ ASC Antibody secreting cell ICOS Inducible co-stimulator ICOSL Inducible costimulatory ligand GC-TFH Germinal center- T follicular helper

IL Interleukin

DC Dendritic cell Bcl6 B cell lymphoma 6 T-bet T-box transcription factor PSGL-1 P selectin glycoligand-1 Ly6c Lymphocyte antigen 6 complex Tcm T central memory

PD-1 Programmed cell death protein 1 NAD Nicotinamide adenine dinucleotide FR4 Folate receptor 4

LCMV Lymphocytic choriomeningitis virus TRM T resident memory

FRT Female reproductive tract

LN Lymph node

Tem T effector memory Klf2 Krüppel-like factor 2

S1pr1 Sphingosine-1-phosphate receptor 1 P2X7R P2X purinoceptor 7

Ccr7 C-C chemokine receptor type 7 Tcf7 Transcription factor 7

KO Knockout

(7)

iBALT Inducible bronchus associated lymphoid tissue HSV Herpes Simplex Virus

SARS-CoV Severe Acute Respiratory Syndrome- Coronavirus CNS Central Nervous System

IBD Inflammatory Bowel Disease IAV Influenza A Virus

HA Hemagluttinin

NA Neuraminidase

ARDS Acute Respiratory Distress Syndrome PR8 Strain A/Puerto Rico/8/34 (H1N1) X31 Strain A/X-31 (H3N2)

NS-1 Non-structural protein 1 M1, M2 Matrix protein 1, 2

PA Polymerase acidic protein Treg Regulatory T cell

HIF-1α Hypoxia Inducible Factor 1A VHL Von Hippel-Lindau

μMT IgM μ-chain mutant OAS Original antigenic sin COVID Corona virus disease RNA Ribonucleic acid Fth1 Ferritin heavy chain 1

Tnfsf4 Tumor necrosis factor superfamily 4 Eif1 Eukaryotic Translation Initiation Factor 1 Id3 Inhibitor of DNA binding 3

mTOR Mammalian/Mechanistic Target of Rapamycin TRAF Tumor necrosis factor receptor-associated factor AREG Amphiregulin

EGF Epidermal Growth Factor

LCMV Lymphocytic Choriomeningitis Virus

(8)

Summary

Heterogeneity is the hallmark feature of CD4 T cells with the ability to differentiate into effector cells of various phenotypes and perform distinct function. This diversification that is essential for optimal response has not been described in the tissue. Using influenza as an infection model, we investigated long-lived antigen-specific CD4 T cells in the tissue. We identified a heterogeneous population of T resident memory cells (TRM) in the tissue, specifically, a T follicular helper-like subset we called T resident helper (TRH) that was previously not reported.

Further characterization by single cell RNA sequencing analysis revealed that TRH possess a unique tissue-specific transcriptional signature distinct from lymphoid TFH and that they can be generated independently of lymphoid replenishment. Histological analysis showed that TRH cluster closely with B cells in inducible bronchus-associated lymphoid tissue (iBALT) and require continued BCL6 signaling for their localization. Interestingly, at very late time points, signals through cognate antigen presentation were dispensable for TRH maintenance and iBALT integrity. Upon heterologous challenge, TRH cells support local antibody production highlighting a previously unexplored function of TRM.

(9)

1. Introduction

1.1 Immunology and Vaccines

Vaccines prime the immune system to elicit a protective response against the invading pathogen. The practice of vaccination began in ancient times with the crust of smallpox virus being introduced into open wounds. This resulted in mild symptoms that soon passed, and the individual was subsequently protected from reinfection with smallpox. Knowledge of the method spread to Europe in the 18^th century and the practice of vaccination acquired scientific status after Edward Jenner’s first vaccination of a child with cowpox, a milder virus than smallpox. With vaccination, the deadly smallpox was eradicated although the mechanism by which protection was conferred was still unknown. Empirically developed undefined vaccine preparations pose a major safety concern. The need to elicit a predictable immune response necessitated a shift from empirical to rational vaccine design ^{1, 2}

Modern vaccination utilizes whole-killed or live-attenuated pathogens that are killed or weaker versions of the pathogen to elicit an immune response without causing a full disease. Subunit or recombinant vaccines prime the immune system with specific pieces of the pathogen or DNA. Antigen-adjuvant combinations were explored to engage the innate immune system that in turn triggers an adaptive immune response. Although the primary line of defense after pathogen entry is a rapid, non-specific response by the innate immune system, adaptive immune responses shape long-lasting immunity against future infections.

In the modern era, pathogens like Mycobacterium tuberculosis (M.tb), HIV-1, influenza and malaria still pose a threat to humankind with their ability to mutate or otherwise evade the immune system.1 A deeper understanding of various pathogens and the variability in infectivity made clear that a one-size-fits-all vaccine is not sufficient.

(10)

Current vaccines primarily aim to promote neutralizing antibody production by B cells, but a robust antibody response alone is insufficient to provide long-lasting protection. Increasing knowledge about cells that aid antibody generation or promote secondary B cell responses highlight the possibility of harnessing T cell immunity as a vaccine target. T cells are found at mucosal surfaces which is the entry site of pathogens. T cells generated at one site with a priming immunization can be recruited to another by a challenge dose through the prime-and- pull strategy.¹ T follicular helper cells (TFH) support B cell growth, proliferation, differentiation and support germinal center (GC) reactions required for formation of antibodies.³ TFH cells are used as a readout of successful vaccination^4,5,6. However, the reported correlates are from TFH cells in circulation whose phenotypes don’t mirror those in germinal centers^7,8,9. Moreover, the contribution of TFH to mucosal immunity has not been widely explored.

1.2 CD4 T cell memory: friend or foe?

1.2.1 CD4 T cell differentiation

Naive T cells survey the body through blood and lymphatics. When a micro-organism invades the host, antigen-presenting cells (APC) process foreign antigens and present them as peptides loaded on Major histocompatibility complex (MHC)-II complexes to the T cell receptor (TCR). Signals received through TCR, co-stimulatory receptors and extrinsic cytokines trigger CD4 T cell activation and differentiation into heterogeneous effectors of unique phenotype and function. CD4 T cells undergo rapid proliferation at the peak of infection response. Once the infection is resolved, many of the effector cells die and some CD4 T cells persist as memory cells in circulation, secondary lymphoid organs and tissue. Memory cells possess the capacity to rapidly respond upon re-infection.¹⁰

(11)

1.2.2 CD4 T cell response

T helper 1 (Th1) and TFH cells are generated upon viral infections. Th1 cells migrate to tissues and through IFN-γ signals, activate innate cells and CD8 T cells for heightened response and accelerated clearance of infected cells.¹¹ TFH cells express chemokine receptor CXCR5 and respond to CXCL13 chemokine cues to navigate into GCs present in secondary lymphoid organs (SLO). Germinal centers are complex structures formed in SLO in response to T cell dependent antigen. Supported by a network of follicular dendritic cells, the GC is characterized by B cell follicles separated by a T cell rich area¹². Here, TFH cells provide survival signals to some of the B cells that are undergoing somatic hypermutation and class-switching to differentiate into high affinity antibody secreting cells (ASC) with specialized effector function.

These antibodies are essential for viral control and protection against future infection.¹³

1.2.3 Signals for Th1/TFH generation, function and memory

The signals through the TCR, co-stimulatory molecules and cytokines determine the differentiation of naïve CD4 T cells to Th1 or TFH fate, function and maintenance at memory phase.

TCR

Several contrasting studies outline the delicate balance of antigen affinity, dose and dwell- time that determine T cell fate decision ^14,15,16. The effect of TCR on GC-TFH is unique in that the APC-T-cell interaction is further supported by ICOS-ICOSL interactions.¹⁷ The nature of these interactions is one of the factors that influences the longevity of T cells in their niche and their phenotype and function upon re-challenge.

Co-stimulation

ICOS-ICOSL interactions are indispensable for differentiation into TFH cells. Some models outline a B cell independent promotion of TFH fate^18,19 while others implicate B cells in the

(12)

initial priming for TFH fate decision.^20,21 The Crotty group describe a possible transition of antigen presentation from DC to B cells during TFH formation. During early priming, pre-TFH can be generated without B cell help but later TFH differentiation is impaired by B cell deficiency²², as also demonstrated by the Jenkins group²³. Throughout the TFH differentiation process, ICOS-ICOSL interactions are crucial, first provided by DC, then by B cells. B cells are dispensable, however, for differentiation into Th1.

Cytokines

The timing, amount and ratio of inflammatory cytokines drive specific T cell fates. IL-6 and IL- 21 are associated with TFH differentiation while IL-12 supports an eventual Th1 fate.²⁴ Early IL-2 producers were shown to differentiate into TFH while the IL-2 consumers followed a path to Th1 differentiation¹⁹. The existing cytokine milieu might be a key determinant in predicting cell fate outcome.

Transcription factors

T-BET and BCL6 are the key transcription factors implicated in regulating Th1 and TFH cell fates respectively although they are not mutually exclusive. Their relative expression can promote or inhibit genes responsible for cell fate decision. In recent studies, BCL6 was highlighted to be a “repressor of repressors” in TFH including repression of itself, highlighting delicate autoregulation. Functional complexes of T-BET-BCL6 are also described in TFH cells.

Key Th1 factors are regulated by T-BET in TFH cells. The permissive state of Bcl6 and T-bet loci point to a state receptive to environmental cues from changing infection settings.²⁵

CD4 T cell memory

Long-lived CD4 memory have been classified into 3 subsets: PSGL-1^hiLY6C^hiTh1 memory, PSGL-1^hi LY6C^lo T central memory (Tcm) and more recently PSGL-1^lo LY6C^lo TFH memory each with their distinct phenotype, migration and recall properties.²⁶ CXCR5⁺PD-1⁺ TFH effector cells were earlier thought to be absent among T cells surviving long term after the

(13)

effector phase.²³ A recent discovery highlighted that TFH cells are susceptible to NAD-induced cell death during cellular processing²⁶. This phenomenon has long biased the conclusions about the longevity of TFH cells. Additionally, Folate receptor 4 (FR4) was described as a phenotypic marker that better discriminates TFH memory cells from Tcm. TFH memory cells can be detected up to 400 days after viral infection, support long-term maintenance of plasma cells in the spleen and display phenotypic plasticity upon recall.²⁶

CD4 T cell memory has been shown to correlate with protection upon secondary challenge infections.²⁷ But specifically eliciting a CD4 T cell memory response as a vaccination strategy has been an interesting research topic. Work by Penaloza-MacMaster et al. showed that specifically re-activating CD4 memory T cells in a chronic LCMV model resulted in excessive weight loss, cytokine storm and multi-organ failure. Microarray analysis revealed a lack of exhaustion markers and a cytokine signature indicative of a bias toward a Th1 phenotype.²⁸ Similarly, dysregulated TFH cells were implicated in pathogenesis of autoimmune diseases and allergy.^29,30 In contrast, a study of antigen specific T cells against SARS showed that re- activated memory CD4 T cells producing IFN-γ upon recall correlated with protection. Although the focus of the publication was on the Th1-like subset, other parenchymal CD27^hi IFN-γ^lo T cell subsets were not absent in the recall response.³¹ These studies highlight an important role for the heterogeneity of CD4 T cell response in balancing host protection and pathology.

(14)

Fig1: CD4 T cell response during infection

(15)

2. Immunological memory in barrier tissues

The understanding of immunological memory has changed over time. Earlier thought to be maintained in blood and secondary lymphoid organs, a new wave of studies has shown that memory T cells can be maintained in barrier tissues. These cells called T resident memory (TRM) are phenotypically and transcriptionally distinct from their circulating memory counterparts. Their presence in mucosal niches makes them the first responders to tissue- specific infection. This feature makes TRM a promising potential vaccine target.^32,33,34

2.1 TRM

2.1.1 Identification

TRM were first identified in tissue isolates after infection that retained their function ex- vivo.^35,36,37 Parabiosis experiments^38,39,40, confocal microscopy⁴¹, intravital imaging⁴² and intravascular staining⁴³ confirmed their tissue-resident property. The generation of antigen- specific CD4 and CD8 TRM were identified in an influenza infection model.⁴⁴ These cells were enriched in the mucosal tissue, exhibited tissue tropism on transfer to naive hosts and displayed unique surface marker expression patterns compared to their circulating counterparts³⁸. Over the years, TRM were identified in lung, liver, kidney, FRT, intestine and skin in various infection models with CD4 TRM outnumbering CD8 TRM in barrier tissues⁴⁵. In humans, endogenous TRM were also identified in tissue transplants up to one year in lung⁴⁶ and after longer durations in intestinal grafts.⁴⁷

2.1.2 Circulation ability

The circulatory capacity of TRM has long been under scrutiny. Early experiments confirmed that TRM are not re-circulating effector memory cells.^32,33 Although parabiosis experiments confirmed the non-circulating nature of TRM, time-course studies after influenza showed attrition of CD8 TRM.⁴⁸ Integrated stress responses and amino acid starvation promoted the

(16)

apoptosis of CD8 TRM, preventing their long-term retention.⁴⁹ This phenomenon occurred in the lung but not in other barrier tissues. At late time points after influenza infection, retrograde migration of TRM to lung draining LN was also observed. These cells retained a TRM phenotype but were found in the LN-associated lymphatics.⁵⁰

2.1.3 Phenotype

Bulk sequencing analysis of antigen-specific memory CD8 T cells across tissues confirmed that TRM are transcriptionally distinct from Tem and Tcm.⁵¹ TRM show downregulation of Klf2 and S1pr1 genes and upregulation of Cd69 which are important for cell retention in tissue.

TRM express chemokine and integrin molecules essential for homing to and binding tissue surfaces. TRM share a core transcriptional signature that is conserved between mice and humans.⁵² Some TRM markers are also shared between CD4 and CD8 TRM: PD-1, CD69, CXCR6. Genes associated with TCR signaling and cytokine production are also enriched in TRM indicating a state of heightened activation poised for rapid response.⁴⁵ P2X7R is enriched on CD8 TRM and TFH memory cells possibly mirroring the metabolic status of cells

“resident” in their respective niches.⁵³

2.1.4 Factors responsible for TRM generation and maintenance

Antigen is required for the generation of T cells in the tissue with ‘classical’ TRM phenotype.³³

Cognate antigen recognition by CD4 T cells at the effector phase is required for TRM formation.⁵⁴ CD8 and CD4 TRM can survive in naïve hosts in an antigen and inflammation- independent manner^50,38,55 although their long-term maintenance is negatively impacted and accompanied by phenotype changes.⁵⁵

Cytokines IL-2, IL-15 and IL-7 are implicated in TRM generation.^32,33 IL2 is required for Th1 and Th2 TRM accumulation.^56,57,58 IL-7 is required for the generation of both CD4 and CD8 TRM^32,33 although IL-15 is dispensable for CD4 TRM.⁴⁵

(17)

Among CD4 TRM, until recently, only Th1 associated cytokine production has been reported.^59,60

Many transcription factors are described to define tissue residency, but most have been associated with CD8 TRM.⁴⁵ HOBIT and BLIMP1 in CD8 TRM were shown to bind S1pr1, Ccr7 and Tcf7 genes associated with tissue egress.⁵¹ In CD4 TRM, Hobit and Blimp1 knockout (KO) show decreased TRM accumulation.⁶¹ Bhlhe40, which was previously implicated in maintenance of mitochondrial fitness in CD8 TRM,⁶² has been shown to be essential for CD4 TRM sustenance in the tissue.⁶³

Intercellular interactions are crucial for TRM maintenance. Infection with influenza has been

shown to generate inducible bronchus associated lymphoid tissue (iBALT). These structures closely resemble germinal centers in secondary lymphoid organs.⁶⁴ iBALT can be formed independently of secondary lymphoid organs and can sustain TRM responses without the need for lymphoid replenishment.⁶⁵ Cytokine and chemokine signals provided by cells within iBALT or cell clusters in the tissue can recruit immune cells and sustain these structures. The presence of iBALT also correlate with reduced pathology in allergy models.⁶⁶ iBALT destruction by DC depletion in the lung is associated with significant decrease in local and systemic antibody production upon re-challenge.⁶⁷ These studies highlight the importance of iBALT in preserving immune homeostasis.

2.1.5 TRM function: protective and pathogenic roles

Upon activation, TRM proliferate locally, produce cytokines, recruit innate cells and dominate the local secondary response.^68,69,70. Secondary effectors of CD4 TRM display a multi cytokine production profile compared to primary effectors. Here TFH effectors were identified although their tissue residency status was not clearly defined.⁵⁹ CD4 TRM have been associated with protection in infection models like influenza⁷⁰, M.tb⁵⁵, Herpes Simplex Virus (HSV)⁷¹, and

(18)

Severe Airway Respiratory Syndrome coronavirus (SARS-CoV)³¹, but were shown to be pathogenic in the context of asthma⁷², psoriasis⁷³, CNS autoimmunity⁷⁴ and IBD⁷⁵. When focusing on eliciting TRM response as a vaccine target, the immune context needs to be considered.

Fig2: CD4 TRM: knowns and unknowns

(19)

2.2 Influenza and CD4 TRM

2.2.1 Influenza virus

Influenza is a member of the Orthomyxoviridae family. It contains a segmented, negative sense, single stranded RNA genome. Influenza falls into 4 genera A, B, C, D and can infect a wide spectrum of species. Influenza A virus (IAV) is classified on the molecular structure and genetic characteristics of Hemagluttinin (HA) and Neuraminidase (NA)⁷⁶. IAV infects airway cells causing alveolar epithelial injury and failure of gas exchange. Extreme consequences are acute respiratory distress syndrome (ARDS) and death. IAV is also the causative agent for the pandemic and circulating seasonal epidemic⁷⁷. Annual epidemics result in worldwide deaths ranging between 290,000-650,000.⁷⁸

2.2.2 IAV proteins as immune targets

Nucleoprotein (NP) of the influenza virus is the key component of the ribonucleoprotein

(RNP) complex and is required for viral transcription and replication. NP is a conserved CD4 and CD8 T cell epitope. Mouse adapted strains PR8 (H1N1) and H3N2 (X31) are the most commonly used strains administered in succession for the study of heterologous infection models. Here, the NP-epitope overlaps between infections and NP-specific memory cells are re-activated to launch the secondary response ^76,79. Focus on generating T cell memory against conserved epitopes may be a good strategy for the design of a universal vaccine.

Hemagluttinin is the surface glycoprotein of the flu virus that is important for attachment to

host, cell fusion and viral entry. HA contains epitopes that trigger neutralizing antibody production and is the key target for the seasonal influenza vaccine. HA is the dominant determinant that triggers viral mutation and recombination. Mutations in HA lead to rapid viral escape that renders neutralizing antibodies to previous infection ineffective, necessitating new vaccine formulations annually. ^76,79

(20)

Neuraminidase (NA), Non-structural protein 1 (NS1), Matrix proteins 1 and 2 (M1, M2) and Polymerase A (PA) are other epitopes being considered as vaccine candidates. These epitopes cannot independently elicit a strong immune response or generate long-lived memory.^76,79

2.2.3 Open questions and goals of thesis

Previous reports of CD4 TRM generated in response to IAV describe a Th1 phenotype.38,71,80,59

Previously, eliciting a Th1-biased memory response as a vaccine strategy has been shown to be detrimental to the host.²⁸ In an asthma model, decreased accumulation of CXCR5-negative T cells correlated with less-severe host pathology, highlighting that an inflammatory cytokine- mediated response is not the only form of protection.⁵⁷ Bulk sequencing of T cells in tissues to define a residency signature creates a bias towards cytotoxic or IFN-γ-producing cells due to their predominance during immune response, thus obscuring T cell heterogeneity.⁷⁰ Transgenic mouse models have also been shown to display impaired generation of heterogeneity unlike their polyclonal models, leading to misinterpretation of the immune setting.²⁶ The classical polyclonal CD4 T cell diversification upon infection described earlier has not been reported in the resident compartment.

3. Aim

To examine the hallmark heterogeneity of CD4 T cells in the tissue resident compartment and refine T cell residency signature to account for heterogeneity.

(21)

4. Results

Swarnalekha et al., Sci. Immunol. 6, eabb6808 (2021) 8 January 2021

S C I E N C E I M M U N O L O G Y | R E S E A R C H A R T I C L E

1 of 16 A N T I V I R A L I M M U N I T Y

T resident helper cells promote humoral responses in the lung

Nivedya Swarnalekha¹*, David Schreiner¹*, Ludivine C. Litzler¹,

Saadia Iftikhar², Daniel Kirchmeier¹, Marco Künzli¹, Young Min Son^3,4, Jie Sun^3,4, Etori Aguiar Moreira⁵, Carolyn G. King^1†

Influenza is a deadly and costly infectious disease, even during flu seasons when an effective vaccine has been developed. To improve vaccines against respiratory viruses, a better understanding of the immune response at the site of infection is crucial. After influenza infection, clonally expanded T cells take up permanent residence in the lung, poised to rapidly respond to subsequent infection. Here, we characterized the dynamics and transcrip- tional regulation of lung-resident CD4⁺ T cells during influenza infection and identified a long-lived, Bcl6-dependent population that we have termed T resident helper (TRH) cells. TRH cells arise in the lung independently of lymph node T follicular helper cells but are dependent on B cells, with which they tightly colocalize in inducible bronchus-associated lymphoid tissue (iBALT). Deletion of Bcl6 in CD4⁺ T cells before heterotypic challenge infec- tion resulted in redistribution of CD4⁺ T cells outside of iBALT areas and impaired local antibody production. These results highlight iBALT as a homeostatic niche for TRH cells and advocate for vaccination strategies that induce TRH

cells in the lung.

INTRODUCTION

Seasonal influenza epidemics are a major cause of global morbidity and mortality. Although annually administered influenza vaccines are among the most widely used in the world, vaccine-elicited neutralizing antibodies offer poor protection against different influenza strains (1). In contrast, there is evidence that prior influenza infection can accelerate viral clearance after heterotypic infection in both mice and humans (2–6). Emerging data suggest that the targeted generation of CD4⁺ memory T cells recognizing conserved epitopes from inter- nal viral proteins may form the basis of a universal influenza virus vaccine (7, 8). CD4⁺ memory T cells are induced after immunization or infection and can be recalled to generate secondary effectors during a challenge infection. Several subsets of CD4⁺ memory cells have been described, including central memory and effector memory cells, which circulate through secondary lymphoid tissues (LTs) and nonlymphoid tissues (NLTs) (9). More recently, T resident memory (TRM) cells that persist in barrier tissues such as lung and skin have been described (10). Although CD4⁺ T cells actually outnumber CD8⁺ T cells in barrier tissues, most of the studies have focused on the requirements for CD8⁺ TRM cell differentiation. In addition, although CD4⁺ T cells are renowned for their substantial plasticity during immune responses, less is known about diversification within the CD4⁺ TRM cell compartment (11–16).

Influenza infection induces the differentiation of CD4⁺ TRM cells in the lung, where they are maintained in an antigen and inflammation- independent manner (17). After a lethal rechallenge, influenza-specific CD4⁺ TRM cells rapidly produce effector cytokines and promote both viral clearance and host survival (7). Lung CD4⁺ TRM cells can also

be induced by mucosal vaccination and were shown to mediate superior protection after heterologous infection, highlighting their potential as a universal vaccine target (18). Influenza-specific CD4⁺ TRM cells are generally characterized as T helper 1 (TH1)–like, with the capacity to produce both interferon-g (IFN-g) and interleukin-2 (IL-2) (7, 19). However, IL-2–deficient memory CD4⁺ T cells were recently shown to provide superior protection compared with wild- type memory cells, an outcome that correlated with decreased inflammation and host pathology during rechallenge (20). These data suggest that protection mediated by CD4⁺ TRM cells is not strictly dependent on their ability to produce effector cytokines and that a balanced secondary response is likely to involve the recruitment and coordination of distinct and specialized CD4⁺ TRM cell subsets (21).

Heterogeneity within the non–antigen-specific CD4⁺ TRM cell compartment was recently examined in a study that reported enrichment of genes associated with the tumor necrosis factor (TNF) receptor superfamily and nuclear factor kB pathways in barrier T cells compared with T cells isolated from their respective draining lymphoid compartments (11). The residency signature derived from this dataset, however, does not take into consideration differences between distinct T cell subsets. Accordingly, this approach tends to overrepresent genes associated with type 1 helper T cells such as Klrg1, Itgae, Id2, and Cxcr6, which may incorrectly define residency for other TH cell subsets. Consistent with this idea, the authors reported the presence of a stronger TRM phenotype in lymphoid TH1 memory cells compared with lymphoid T follicular helper (TFH) cells, which they attributed to the relative ability of these cells to adapt to barrier tissues. Similarly, another study addressing the relation- ship between TRM cells present in secondary lymphoid organs and TFH memory cells reported distinct transcriptional profiles of these subsets (12). However, in this case, the authors analyzed T cell receptor (TCR) transgenic cells that have impaired TFH memory cell generation relative to polyclonal antigen–specific cells (22), likely enhancing a bias in TRM toward genes expressed by cytotoxic and type 1 cytokine- producing cells. On the other hand, it is well appreciated that TFH cells share many surface markers and molecular dependencies

1Immune Cell Biology Laboratory, Department of Biomedicine, University of Basel, University Hospital Basel, CH-4031 Basel, Switzerland. ²Personalised Health Basel- Oncology Cluster Basel, University of Basel, Basel, Switzerland. ³Department of Medi- cine, Mayo Clinic, Rochester, MN 55905, USA. ⁴Department of Immunology, Mayo Clinic, Rochester, MN 55905, USA. ⁵Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland.

*These authors contributed equally to this work.

†Corresponding author. Email: carolyn.king@unibas.ch

exclusive licensee American Association for the Advancement of Science. No claim to original U.S.

Government Works

at OFFENTLICHE BIB DER UNIV BASEL on January 12, 2021http://immunology.sciencemag.org/Downloaded from

(22)

2 of 16

with TRM cells, including high expression of programmed cell death protein 1 (PD1), P2X7R, CD69, and inducible T-cell costimulator (ICOS), and a requirement for S1PR1 and KLF2 down-regulation to develop (10, 23–26). In support of this idea, TFH memory cells isolated from the spleen after lymphocytic choriomeningitis virus (LCMV) infection display a partially overlapping transcriptional signature with TRM cells, consistent with the noncirculating nature of both subsets (22). However, whether different tissue-resident TH cell subsets express conserved or distinct residency signatures compared with their lymphoid counterparts remains unresolved.

In this study, we characterized the dynamics and diversification of polyclonal CD4⁺ T cells present in the lung and draining lymph node (dLN) after influenza, using the resolution provided by single-cell RNA sequencing (scRNA-seq) to decouple residency and functional signatures between these two tissue compartments. Our analyses of influenza-specific CD4⁺ T cells reveal notable heterogeneity within the lung compartment, composed of two broad subsets that we designate T resident helper (TRH) and TRM1 cells. TRH cells persist stably and depend on expression of the transcription factor (TF) BCL6 for their differentiation. TRH cells are tightly colocalized with B cells in the lung and require ongoing antigen presentation for their maintenance. CD4⁺ T cell–intrinsic deletion of Bcl6 at late time points after primary infection impairs the local humoral response upon reinfection. These data identify a TRH subset that may be a rational target to drive potent and protective immunity in the lung mucosa.

RESULTS

Transcriptional residency signature in CD4⁺ T cells is biased toward a TH1 phenotype

To examine influenza-specific CD4⁺ memory T cell populations, we performed scRNA-seq on memory CD4⁺ T cells from the lung and lung-draining mediastinal LN of mice 30 days after infection with the PR8 strain of influenza. Virus-specific T cells were detected by tetramer staining for IA^b:NP^311-325 and were largely protected from intravascular antibody staining for CD45 (Fig. 1A and fig. S1A), a method confirmed to identify tissue-resident cells by parabiosis experiments (27). Treatment of mice with fingolimod (FTY720) to block the egress of memory cells from lymphoid organs did not alter the total number of NP⁺ CD4⁺ T cells isolated from the lung, indicating that this population can be sustained without input from the circulation at this time point (fig. S1, B and C). After quality control, normalization, and dimensionality reduction, the tran- scriptomes of lung and LN T cells were clearly distinct, and an ex- pected small population of putative circulating cells identified by elevated expression of S1pr1, Il7r, and Bcl2 was also detected in the lung (Fig. 1, B and C, and table S3). Differential expression analysis between lung and LN cells confirmed enrichment of previously reported CD4⁺ residency genes (e.g., Crem, Vps37b, Rora, Ramp3, and Tnfrsf18) in the lung (Fig. 1C) (11, 12). To determine whether this residency signature was conserved in another viral infection model, we analyzed GP66-specific memory T cells from the liver and spleen of LCMV-infected mice at 37 days after infection (Fig. 1D).

Lung-specific genes from influenza were broadly enriched in liver cells from LCMV and vice versa (fig. S1D). To better define a conserved residency signature, we examined the intersection of genes enriched in LTs and NLTs from both viral infections (Fig. 1E). This tissue-level comparison demonstrated an apparent bias in NLT to-

ward a TH1 phenotype, with Ccl5, Crip1, Vim, Nkg7, Id2, and Cxcr6 among the top shared genes (Fig. 1, E and F). Consistent with transcriptional similarities between the type I cytokine-producing CD4⁺ and CD8⁺ cells, lung and liver cells also showed enrichment for a multitissue signature for CD8⁺ TRM (Fig. 1G). Conversely, TFH signatures including genes such as Tcf7, Izumo1r, Dennd2d, Shisa5, Rgs10, and Cxcr5 were enriched in LT, in line with the conception of TFH as a resident population of secondary lymphoid organs (Fig. 1, E and F) (23). The presence of TH subset–specific genes in both NLT and LT cells reported here and elsewhere calls for a residency signature that separates function from niche (11, 12).

Transcriptional heterogeneity in the lung reveals a tissue signature distinct from T helper function

We next investigated heterogeneity within the NP-specific CD4⁺ T cell compartment at day 30 in the lung. Unsupervised hierarchical clustering classified the cells into four populations (Fig. 2A). We found clear evidence of a TFH-like phenotype in the lung: cluster 3 showed increased expression of Sostdc1, Sh2d1a, Ppp1r14b, Rgs10, Id3, and Tcf7 (Fig. 2B and table S3). Although this TFH-like cluster was lower in many of the genes characterizing residency compared with the other lung clusters, expression was still clearly higher compared with cells in the LN (fig. S2A). Cells in cluster 2 had a TH1 phenotype including enrichment for Selplg, Nkg7, Ccl5, Id2, and Cxcr6. The similarity of these clusters to known T helper subsets was further confirmed by scoring each cell according to published TFH and TH1 memory gene sets (Fig. 2C). Elevated transcription of S1pr1, Il7r, and Bcl2 suggested that cluster 4, distinct in principal component 2 and further distinguished by high ribosomal protein content, contained circulating cells recovered during lung processing. While cluster 1 was most similar to the TFH-like cluster, it had a distinct profile characterized by enrichment for Hif1a, Areg, Tnfrsf4, and Tnfsf8 and also expressed more Nr4a3, suggesting ongoing major histocompatibility complex II (MHC-II) engagement and activation of the calcineurin/nuclear factor of activated T cell pathway (Fig. 2, A to C, and fig. S2B) (28). Cells in cluster 1 were also enriched for Rora, which is reportedly induced by extrinsic signals in the micro- environment and plays a role in negatively regulating lung inflammation during infection (29). Further examination of the differences between these two TFH-like clusters was reserved for future investigation.

We next focused our analysis on NP-specific CD4⁺ T cells from the dLN at day 30, cutting the resulting hierarchical clustering tree to four clusters (Fig. 2D). These clusters were classified similarly to those found in the lung, with a small TH1 cluster 4 expressing Nkg7, Ccl5, Id2, and Cxcr6 and tracking with ribosomal protein-rich cluster 1 along principal component 1 (PC 1); strongly TFH cluster 3 with high expression of Pdcd1, Sh2d1a, and Sostdc1; and TFH-like cluster 2 enriched for Izumo1r, Tox, and Hif1a (fig. S2C and table S3). As with the lung data, we scored LN cells using TFH and TH1 gene sets to further verify this classification (fig. S2D). To generate a more representative signature for lung residency, we performed differential expression analysis on each of the most phenotypically similar cluster pairs between lung and LN, for example, comparing LN TFH with TFH-like lung cells and LN TH1 with TH1-like lung cells.

The final conserved residency signature consisted only of genes that were differentially enriched in all of the noncirculating lung clusters compared with their LN counterparts (Fig. 2E). This enrichment was confirmed in our LCMV dataset and other recently published TRM data (fig. S2, E to G). Subsequent removal of these conserved

(23)

3 of 16

residency genes allowed us to identify TH subset–specific genes differentially expressed between tissue and lymphoid compartments.

T_H1-like cells in the lung were enriched over their LN counterparts for genes including Ctla2a, Crip1, Lgals1, Tnfrsf18, and Infgr1, whereas TFH-like lung cells exhibited higher expression of Mt1, Hspa1a, Klf6, Tnfsf8, and Areg (Fig. 2, F and G, and table S3). In addition, we compared the two TFH-like clusters in the lung and found a more

lymphoid signature in cluster 3 and enrichment of the residency signature in cluster 1, suggesting functional diversification of TFH-like cells in the lung (Fig. 2H and table S3).

To gain further insight into the transcriptional regulation of distinct resident CD4⁺ T cell subsets, we next assessed TF activity using single-cell regulatory network inference and clustering (SCENIC) (30). This approach identified active regulatory activity in TFs such as hypoxia inducible factor 1 (HIF-1A), CAMP responsive element modulator (CREM), Fos-related antigen 2 (FOSL2), and known T_H1-related factors PR domain zinc finger protein 1 (PRDM1), Runt- related transcription factor 2 (RUNX2), and Runt- related transcription factor 3 (RUNX3) (fig. S2H). Cluster-specific analysis revealed that the activity of PRDM1, RUNX2, and RUNX3 was limited to TH1-like clusters, whereas HIF-1a activity was focused primarily in lung cluster 1, mirroring gene expression of Hif1a. Bcl6 regulatory activity was detected in lung TFH-like cluster 3 and LN TFH cluster 3, along with CCAAT/

enhancer-binding protein alpha (CEBPA), which is reported to inhibit both IFN-g production and TCR-driven proliferation (fig. S2I) (31, 32).

FOXP3- associated regulatory activity was enriched in lung over LN although there was only minimal expression of Foxp3 transcript in the dataset and FOXP3 cells were not detected by fluorescence-activated A

CD44

IA^bNP

Naïve Day 30

20.5 0.4

CD44

IA^bNP

Naïve Day 30

Lung Lymph node

B C

D ^{PC 1}

PC 2

Circulating (n = 224) LN (n = 1412) Lung (n = 2556)

Circulatin g

LN Lung

−2 −1 0 1 2

Flu day 30 Flu day 30

0.0

0 103104105 0

103 104 105

0 103104105 0

103 104

105 0.0

Crip1 Bcl2Il7r Rps19 Rpl13aRps5 Rpl32 Ifngr1 Rpl14Emb Rplp0 Rps16Rps7 Rpl35 Gramd3 Rpl10a Rpl18a Rps4x Rps18Rpl8 Ppp1r14b Rps29Ltb mt−Nd3Chd3 Limd2 PtprcapIl16 Rps28 Apobec3 Dennd2d Gm10263 Tspan32 Rps27rtTcf7 Gm42418 Asap1 Eif3f Lrrc58 Rpl38 Tnfrsf4 Vps37bCrem Nkg7Cstb Hspa1a Ccl5Areg H2afz Ctla2aSub1 Ramp3 Tnfrsf18Fth1 H3f3bMt1 Hif1a Fam107bMmd Hilpda

PC 1

PC 2 ^{Spleen (n}^{= 3488)}Liver (n = 3011)

LCMV day 37

LN flu Spleen LCMV 300

200

100

0

−1 0 1 2

E

LN Lung

Spleen

Liver

0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0

F

TH1

LCMV day 37

Average log2FC NLT vs. LT

−Log10 (adjustedP value)

Gene set average expression Lung flu

Liver LCMV Flu day 30

TFH

0.8 1.2 1.6

0.6 0.9 1.2 NLT

LT

G

Spleen Liver LN

Lung

Gene set average expression 0.2 0.4 0.6

CD8 T_RM signature

0 .

0 0.2 0.4 0.6 0.8

Centered scaled expression

Fig. 1. Inflammatory T cells at site of infection confound a tissue-residency signature. Analysis of scRNA-seq of antigen-specific CD4⁺ T cells >30 days after viral infection. (A to C) PR8 influenza-infected mice 30 days after infection. (A) Gating strategy for NP-specific CD4⁺ T cells in lung and mLN of naïve and infected mice. (B) PCA showing day 30 LN and lung samples, with a putative circulating cluster identified in lung samples. (C) Heatmap showing centered scaled single-cell expression of top 20 genes sorted according to cluster average log2 fold change (FC), adjusted P < 0.05. (D) PCA showing scRNA-seq of GP66-specific CD4⁺ T cells from spleen and liver of mice 37 days after LCMV infection. (E) Differentially expressed genes discriminating lung from LN (flu) or liver from spleen (LCMV). To enable plotting the highest y-axis values, P values of 0 were assigned the lowest nonzero P value. (F) Log-normalized average expression of TFH and Th1 memory signatures (22). (G) Log- normalized average expression of combined CD8 residency signature from multiple tissues (69, 70).

Sequenced cells were pooled from n = 12 mice in (B) and n = 2 in (D).

(24)

4 of 16 cell sorting (FACS) in the antigen- specific

compartment (fig. S2, J and K). In summary, the scRNA-seq data gathered here constitute a comprehensive picture of CD4⁺ T cell memory to influenza and disentangle residency from helper function to highlight not only conserved differences between NLT and LT but also intratissue heterogeneity and functional diversification of lung-resident T cells.

Resident CD4⁺ T cells include TH1-like and TFH-like subsets To determine whether the heterogeneity detected by scRNA-seq was also mirrored at the protein level, we examined the phenotype of NP-specific CD4⁺ T cells using flow cytometry. At day 30 after influenza infection, NP-specific CD4⁺ T cells in the lung could be divided into two major subsets with reciprocal expression of folate receptor 4 (FR4) and P- selectin glycoprotein ligand-1 (PSGL1), markers shown to discriminate virus-

specific TH1 and TFH memory cells. (Fig. 3A) (22). After viral clearance and T cell contraction, the number of cells falling within these two T resident cell subsets remained stable to at least 120 days after infection (Fig. 3B). Treatment with FTY720 at day 30 after infection did not change the number of cells in either subset, indicating that both populations are lung resident (fig. S3A). FR4^hiPSGL1^low

(hereafter TRH) cells expressed many markers associated with TFH

effector cells including PD1, CXCR5, P2X7R, CD73, CXCR4, and ICOS (Fig. 3, C and D). FR4^loPSGL1^hi (hereafter TRM1) cells expressed lower levels of TFH- associated markers but higher levels of markers associated with TH1 cells including CXCR6, T-bet, and CD11a (Fig. 3, E and F). To examine cytokine production by NP-specific

PC 1 tSNE 1

PC 2 tSNE 2

1 (n = 531) 2 (n = 405) 3 (n = 388) 4 (n = 88)

D _{LN day 30}

PC 1 tSNE 1

PC 2 tSNE 2

Gene set average normalized expression TFH memory TH1 memory

1 2 3 4

0.4 0.6 0.8 1.0 1.2 1.4 1

2 3 4

0.5 1.0 1.5 2.0

A B

1 (n = 1279) 2 (n = 699) 3 (n = 504) 4 (n = 298)

C

−2 −1 0 1 2

Lung day 30

Tnfsf8 Tnfrsf4Cstb H2afz Eea1Hif1a Ramp3 PrkcaAreg Ifi27l2a Ramp1Eprs Tbc1d4Nr4a3 Wnk1MmdMt1 Nr4a2 Tnfaip8 Tspan3 Nkg7Ccl5 Plac8 Crip1 S100a6Rgs1 Ctla2a Lgals1 Lgals3 Ifngr1 AW112010 S100a4Cxcr6 Rgs2VimId2 Ms4a4b HilpdaCtsw Hspa1aDnajb1Ccr2 Rgs10 Rnaset2b Sostdc1 Rnaset2aUbac2 2310001H17RikPpp1r14bId3 Sh2d1a PtprcapJun Zfp36Ltb ActbTcf7 Cdk2ap2 Limd2 Rps19Pfn1 Rplp0 Rps5Rps7 Rpl35Il7r Rps20 Rpl13a Gm8730 Rpl32 Rpl10aRps12 Gm2000 Gm10269Rpl12 Gm9493Rps8 Rpl14 Rps4x Rps16

1 2 3 4

E

Lung 3 Lung 1 Lung 2 LN 3 LN 2 LN 4

Crem Eif1Fth1 H3f3b Sh3bgrl3 Sub1Tmsb4x Tnfrsf4 UbbVps37b Anp32b Apobec3 Chd3 Dennd2d Eif3f Gm42418 Il16Limd2 mt-Nd3 Ptpn7 Ptprcap Rps29 Sp100 Tmem160 Tspan32

NLTLT

−2 −1 0 1 2

Average log2FC

−Log10 (adjustedP value)

−Log10 (adjusted P value)

F

0 10 20 30 40 50

−1 0 1 2

5 10 15 20

−1 0 1 2

LN Lung

G H

−1 0 1

Lung 3 Tissue specific Lung 1

0 10 20 30 40 50

Lung 3 Lung 1

LN 4 (TH1) Tissue specific Lung 2 (TH1-like)

LN 3 (TFH) Tissue specific Lung 3 (TFH-like)

T_FH-like cells

Average log2FC Fig. 2. Controlling for T helper function im-

proves tissue residency signature. Analysis of scRNA-seq of NP-specific CD4⁺ T cells at day 30 after influenza infection. (A) Unsupervised hierarchical clustering of lung cells using Ward’s method, visual- ized using PCA and t-distributed stochastic neigh- bor embedding (tSNE). (B) Heatmap showing centered single-cell expression of top 20 cluster- defining genes sorted according to lung cluster average log2FC. (C) Log-normalized average expression of TFH and TH1 memory signatures (22).

(D) Unsupervised hierarchical clustering of LN cells using Ward’s method: PCA and tSNE. (E) Heatmap of centered scaled single-cell expression from both lung and LN showing conserved genes for NLT and LT. Genes included are differentially expressed in the tissue in all three of the noncirculating cluster pairs. (F) Genes discriminating LN TH1 cells from lung TH1-like cells with reference to conserved tissue-specific signature. (G) Genes discriminating LN TFH cells from lung TFH-like cells with reference to conserved tissue-specific signature. (H) Genes discriminating lung TFH-like cluster 3 from lung TFH-like cluster 1 with reference to conserved tissue-specific signature. For (B) and (E) to (H), adjusted P < 0.05. Sequenced cells were pooled from n = 12 mice.